The economic implications of equipment failure are called for effective maintenance
techniques. The research investigates current maintenance practice in several New
Zealand industries and the improvements that could be obtained by the use of
predictive maintenance techniques.
Initial research was undertaken in a series of case studies within New Zealand
industries situated in Auckland. The first two cases studies were of preventative
maintenance techniques of two conveyor lines in a biscuit manufacturing company.
The results showed a well defined preventive maintenance schedules that was
Systems Applications Products (SAP) programme was used to managed for daily,
weekly, monthly and yearly maintenance activities.
A third case study investigated current predictive maintenance technique involving
Fast Fourier Transform analysis of shaft vibration to identify a bearing defect. The
results diagnosed a machine with a ball bearing defect and recommendation was given
to change the bearing immediately and install new one. The machine was opened up,
a big dent was on one of the balls as predicted by the analysis and the bearing was
Research then looked at a novel technique called Cepstrum analysis that al lows the
deconvolution of vibration spectra from separate sources. This allows identification of
several defects from the monitoring of a single vibration signal . Experiments were
carried out to generate transfer functions for different gear faults at two different
loadings. Blind deconvolution of the signal using a homomorphic filter was used to
separate the source forcing frequencies from the structure resonance effects of the two
gear faults, indicating that the technique could be used successfully to monitor
equipment for a range of gear faults occurring simultaneously.